AMD VEGA confirmed for 2017 H1

[JediFragger]

Hold on a second. We are in Q4 2016 not Q1 2017, how Zen is late?
Same applies to Vega. We are still in H2 2016, NOT H1 2017 when Vega comes out.

Nvidia was late with Pascals, from Q1 pushed to Q2 and became available in numbers at Q3 of 2016.
And they are going to be late with Volta. From H1 2017 moved to H1 2018!

AMD originally stated 2016 for Zen, now it's for a 2017 release. Vega was Q1 '17 but has now shifted to H1 '17, which knowing AMD means Computex (ie June).

I am prepared to eat my words, I hope it's far earlier

 

[Dave2150]

Did you not look at the link that Flopper posted ? A very good showing for the Fury X against a card that is a generation in front of it. I am not trolling , I am looking at information and commenting on what I have seen. Nothing more, nothing less.

Oh, we are onto the overclocking thing now are we. You'll be barking on about heat and efficiency next. How about you mention the words "Overclockers Dream" and "AMDs Rubbish Drivers" just to get all the usual constantly regurgitated guff out of the way.

This new generation we only have a small set of cards which are competing right now on the same level and they are the 1060, 1050Ti and 1050 which are competing against the RX480, 470 and 460. I see most of this being neck and neck and tit for tat depending on the games chosen etc...

Vega will release sometime early 2017 and go head to head with 1070/1080, then and only then will we really be able to compare them. So you go and compare Nvidias latest generation to AMDs last generation if it makes you feel better.

Initially, all I said was that the Fury X was seen to beat the 1070 on a specific just released AAA title, that (admittedly) I actually couldn't remember the name of at time of posting, but knew I had seen it. Flopper posted a load of benches that showed the Fury X leading in more than one title. Which until that posting, I wasn't aware of.

You just got yer knickers in a twist because you recently got a 1070 and felt butt hurt. No more, no less.

Recently? I've had a 1070 for many months.

I own several GPU's, including a RX480, 1070, 980M and others. Also a Freesync monitor (XL2730Z) and gsync laptop. Implying I'm some raging fanboy defending his 1070 is just silly.

If you've been around here for any length of time, you'd know I actually prefer AMD hardware. I'm eagerly awaiting the arrival of Vega, which I'll purchase as long as ticks a few boxes.

I should probably stop replying to you as it's simply a waste of my time. Believe what you want to believe, but the Fiji line up of cards was an immense failure - they did not sell well due to many reasons. You'll find they are cheap on OCUK for a reason - hardly anyone wants them.

Here are some roundup benchmarks (covering many popular games)

As we can see, the Fury Pro (air cooled) is around 9% behind the Fury X. The Fury Pro (air cooled) is also barely above NVIDIA's mid range card, the 6GB 1060. Note that the 1060 has more VRAM...

I suggest you and Flopper go and read reputable reviews and benchmarks and educate yourselves on the 1060, 1070 performance at stock and when overclocked (hint, they overclock very well). The air cooled fury is not equal to a 1070.

 

[maonayze]

LOL

How long ago were those benches done. And that is a non specific graph to say the least. We all know that when some review sites do these mass game reviews they have a lot of gameworks titles in them and hardly any of them ever record Doom running on Vulkan, which if taken into account and added into all the other scores would affect an average score round up a fair bit.

As for the failure of Fiji, in my opinion it was an experiment and first dip into the water for HBM which will hopefully bring rewards in the near future (We hope). My Fury Pro is running fantastically well, thanks for asking. I too have an XL2730Z and freesync is a dream to use. No complaints from me.

As for review sites....are there really any reputable ones left in all honesty? They all seem to cherry pick in some way or another and quite a few are now heavily sponsored by companies and given free gear.

If you don't like what I say then put me on your ignore list. Just because I may have a different opinion to you doesnt mean that it is wrong. I may from time to time make mistakes as we all rely on info from various sources which may or may not be reliable and most of which are rumour until a products release.

If I am wrong on something I usually put my hand in the air and apologise for any issue or inaccuracy.

Oh and one last thing....you are right the air cooled Fury isnt equal to the 1070.....I never said it was...I was talking about the Fury X...not sure why you keep mentioning the Fury Pro, as floppers graph points out it is a Fury X NOT a Fury Pro. I do like how you can take an old non specific graph with only 2560x1440 on it and believe it is gospel and you wont even consider Floppers graph which is game specific and has more relevance to the discussion. Now THAT is a "Cherry Pickers Dream".

Have a nice day

 

[JediFragger]

Worst thing about Fury for me was it's non-tweakability

I'm hoping Vega brings it back to the table!!

 

[Bonjour]

Surprised how close the Fury X is to the 980ti in that graph.

 

[Rroff]

Surprised how close the Fury X is to the 980ti in that graph.

Depending how much certain titles weigh in your spread of benchmarks the FX's position can vary a lot - it shows very strong performance with Doom Vulkan for instance and 1-2 other games where its position in the lineup is very different to some other games.

 

[LoadsaMoney]

 

[CAT-THE-FIFTH]

Look,at how the GTX970 4GB thrashes the GTX1060 3GB even though it is meant to have a slower core.

Also,its a Nvidia sponsored game,so even more LOL.

Its even more hilarious when RX470 4GB cards like the Powercolor Devil,actually match the reference RX480 8GB which TPU uses,since it throttles so much.

That would place the card at around RX480 level.

Edit!!

Also,look at that RX470 go!!

Its easy to cherrypick titles if you want to.

 

[Ian Evey]

LOL

How long ago were those benches done. And that is a non specific graph to say the least. We all know that when some review sites do these mass game reviews they have a lot of gameworks titles in them and hardly any of them ever record Doom running on Vulkan, which if taken into account and added into all the other scores would affect an average score round up a fair bit.

As for the failure of Fiji, in my opinion it was an experiment and first dip into the water for HBM which will hopefully bring rewards in the near future (We hope). My Fury Pro is running fantastically well, thanks for asking. I too have an XL2730Z and freesync is a dream to use. No complaints from me.

As for review sites....are there really any reputable ones left in all honesty? They all seem to cherry pick in some way or another and quite a few are now heavily sponsored by companies and given free gear.

If you don't like what I say then put me on your ignore list. Just because I may have a different opinion to you doesnt mean that it is wrong. I may from time to time make mistakes as we all rely on info from various sources which may or may not be reliable and most of which are rumour until a products release.

If I am wrong on something I usually put my hand in the air and apologise for any issue or inaccuracy.

Oh and one last thing....you are right the air cooled Fury isnt equal to the 1070.....I never said it was...I was talking about the Fury X...not sure why you keep mentioning the Fury Pro, as floppers graph points out it is a Fury X NOT a Fury Pro. I do like how you can take an old non specific graph with only 2560x1440 on it and believe it is gospel and you wont even consider Floppers graph which is game specific and has more relevance to the discussion. Now THAT is a "Cherry Pickers Dream".

Have a nice day

Forgive me for butting in but you mention that dave's post is only using a certain resolution, and totally ignore that floppers graph only put the fury x above the 1070 @ 4k by 2%, the rest of the resolutions have the 1070 being on top.

Now THAT is a "Cherry Pickers Dream".

 

[Freddie1980]

Yer, another crappy PC port confirmed .

You have to wonder why console ports run so badly on the PC especially AMD hardware given that the consoles are using the same technology (x86 with GCN).

You would have through an AMD CPU and GPU would be the best combination for console ports on the PC but if you own a FX8350/6300 CPU your more often then not better off buying a mid range Nvidia GPU as there CPU overhead is a lot better then AMD's.

 

[CAT-THE-FIFTH]

You have to wonder why console ports run so badly on the PC especially AMD hardware given that the consoles are using the same technology (x86 with GCN).

You would have through an AMD CPU and GPU would be the best combination for console ports on the PC but if you own a FX8350/6300 CPU your more often then not better off buying a mid range Nvidia GPU as there CPU overhead is a lot better then AMD's.

Depends on the port though.

I still remember the R9 380 thrashing my GTX960 at launch in The Division. That title was Nvidia sponsored too.

Basically some games have the console effect,some are sponsored by AMD,others by Nvidia and you will get games which perform better on one companies hardware than the others.

Then you have a game update or driver update which can suddenly change the relative positions of cards,etc.

Not even AMD or Nvidia sponsoring a game does not predict what performance will be within a few months of launch - look at BF4,Metro2033,GoW remastered,Doom,etc.

 

[Freddie1980]

That's a decent post cat and your right it is game dependant with hits and misses but the trouble with benchmarks graphs for video cards is they only show the 'best case scenario' and are often run with Core i7's/X99 overclocked to 4.5 ghz+. What they never show is the effect of running a more modest CPU such as FX6300 or a Pentium G3258 with mid range video cards. If there is a CPU overhead it will come to the table when running more real world configurations.

 

[humbug]

Depends on the port though.

I still remember the R9 380 thrashing my GTX960 at launch in The Division. That title was Nvidia sponsored too.

Basically some games have the console effect,some are sponsored by AMD,others by Nvidia and you will get games which perform better on one companies hardware than the others.

Then you have a game update or driver update which can suddenly change the relative positions of cards,etc.

Not even AMD or Nvidia sponsoring a game does not predict what performance will be within a few months of launch - look at BF4,Metro2033,GoW remastered,Doom,etc.

It has to be said AMD are doing really well in later games.
Who would have thought the RX 470 would beat a 980TI, a Fury Pro match a GTX 1080.

Amazing performance from AMD.

 

[LoadsaMoney]

Yup, AMD smashing it!

 

[Freddie1980]

It has to be said AMD are doing really well in later games.
Who would have thought the RX 470 would beat a 980TI, a Fury Pro match a GTX 1080.

Amazing performance from AMD.

Providing you running a decent CPU. There's nothing wrong with AMD's video cards pre se but see what happen if you put a more real world CPU with a RX470 or RX370 and compare it with an equivalent Geforce. Google all you want but you will struggle to find benchmarks as most sites use top end overclocked I5's and I7's to test video cards. Richard Leadbetter at Digital Foundry done a video on this very topic, it was an eye opener.

 

[CAT-THE-FIFTH]

That's a decent post cat and your right it is game dependant with hits and misses but the trouble with benchmarks graphs for video cards is they only show the 'best case scenario' and are often run with Core i7's/X99 overclocked to 4.5 ghz+. What they never show is the effect of running a more modest CPU such as FX6300 or a Pentium G3258 with mid range video cards. If there is a CPU overhead it will come to the table when running more real world configurations.

Don't disagree with that either,but I also had a longterm issue with sites not testing games themselves on lower end hardware to see what an entry level gamer needs - will the game run fine on an IGP,or an £80 graphics card paired with a cheaper CPU??

I find YT an increasingly useful way of finding out.

Providing you running a decent CPU. There's nothing wrong with AMD's video cards pre se but see what happen if you put a more real world CPU with a RX470 or RX370 and compare it with an equivalent Geforce. Google all you want but you will struggle to find benchmarks as most sites use top end overclocked I5's and I7's to test video cards. Richard Leadbetter at Digital Foundry done a video on this very topic, it was an eye opener.

Depends on the game though - some games show very little difference. Even seen the odd one where the tables are turned. Again it depends on the game.

Having said that I have tended to err towards Nvidia if someone is using a Core2quad or something similar,but once you get to the territory of IB onwards,etc,I am not so sure.

For instance in some cases it will make zero difference - an RX470 will still thrash a GTX1050TI in almost any game for example. Or a GTX1060 3GB in a game which is more VRAM heavy will still have more issues than a 6GB version or a RX470 4GB/8GB.

Even in the two games I listed the RX470 is doing so well when compared to more expensive Nvidia cards,even lopping off 10% to 15% is not going to change things.

Edit!!

Also I believe there are some improvements to the hw schedular with Polaris too.

Polaris is essentially the same generation as Tonga and Fiji,with some improvements.

You can see an RX480 coming very close to a Fury X at 1080P in those games and faster than a R9 390X too,and I think that hints at some improvements over the older cards in CPU overhead.

Vega is meant to be a bigger change so it will be interesting to see if they have improved in that regard otherwise they will be hitting a much bigger CPU bottleneck with a much faster card.

 

[AllBodies]

Yeah these benches are just madness (in a good way).

RX480 beating the GTX 980 and matching the Fury X in CoD.

And beating the GTX 1070 and R9 390X in Titanfall :O

Wonder what the technical explanation is? Poor optimisation on Nvidia's part, or those games taking advantage of Polaris' upgrades? Or a bit of both.

Either way, things could get very interesting when Vega gets thrown at modern games.

 

[Dave2150]

Yet another game where the 1060 beats the FuryX, hadn't seen these before, thanks for posting.

 

[humbug]

Yeah these benches are just madness (in a good way).

RX480 beating the GTX 980 and matching the Fury X in CoD.

And beating the GTX 1070 and R9 390X in Titanfall :O

Wonder what the technical explanation is? Poor optimisation on Nvidia's part, or those games taking advantage of Polaris' upgrades? Or a bit of both.

Either way, things could get very interesting when Vega gets thrown at modern games.

£250 RX 480 only 10% shy of the £600 GTX 1080 is the kicker

I think its just AMD continuing to improve on software so the hardware has more freedom to stretch its legs.

 

[Panos]

I have to say only this.
From 370.70 drivers onwards, every single NV driver, gives me less performance on benchmarks, having amounted now to -10% between 375.76 and 370.70 on my 1080 @2190

While my Fury Nano, has gained a whooping 20% in benchmark performance between 16.1 and 16.11.3 drivers. And with the later drivers, clocks to 1150 core, best core before was 1140 with 16.9 and before that was capped at 1100.